A NNALES DE L ’I. H. P., SECTION A
D ANIEL M. G REENBERGER
The equivalence principle meets the uncertainty principle
Annales de l’I. H. P., section A, tome 49, n
o3 (1988), p. 307-314
<http://www.numdam.org/item?id=AIHPA_1988__49_3_307_0>
© Gauthier-Villars, 1988, tous droits réservés.
L’accès aux archives de la revue « Annales de l’I. H. P., section A » implique l’accord avec les conditions générales d’utilisation (http://www.numdam.
org/conditions). Toute utilisation commerciale ou impression systématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fichier doit contenir la présente mention de copyright.
Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques
http://www.numdam.org/
307
The Equivalence Principle
Meets the Uncertainty Principle
Daniel M. GREENBERGER (*)
Max Planck Institut fur
Quantcnoptik,
D-8046 Garching b. Munchen, FRGVol. 49, n° 3, 1988, 1 Physique theorique
ABSTRACT. 2014 We
argue
that the very existence of the weakEquivalence Principle
canonly
be understood in terms of theUncertainty Principle.
RESUME. 2014 Nous
presentons
desarguments
montrant que 1’existence meme duPrincipe d’Equivalence
Faible nepeut
etrecomprise qu’en
relation avec Ie
Principe
d’lncertitude.Louis De
Broglie
was, as was AlbertEinstein,
a master iconoclast2014a constant
challenger
ofuniversally
heldmyths.
It is in this
grand,
butdifficult,
tradition ofbeing willing
to step on every- one’s toes, that Jean-PierreVigier
was trained as DeBroglie’s assistant,
and in which he has
fully
matured. Notonly
in hisprofessional
life as aphysicist,
but in hisprivate
life aswell,
has he set his own direction and followed his own conscience.In the same
spirit
I would like to tell youwhy
I think the search for aclassical unified field
theory
failed. I think it alsoexplains why
I expect the search for aquantized
version ofgravity
to also fail. It isprimarily
because I believe the search is
heading
in thetotally
wrong direction. Theproblem
is not the mathematicalcomplexity
of thetheory,
but rather that thephysics
is wrong. I believe that there are still many cluesbeing
offeredby
theequivalence principle,
andinsights
into thisprinciple,
that we havenot made use of. And
they
lead in atotally
different direction.(I recognize
(*) Permanent Address: Dept. of Physics, City College of New York, New York, N. Y.
10031, USA.
Annales de l’Institut Henri Poincaré - Physique theorique - 0246-0211
Vol. 49/88/03/307/ 8 /$ 2,80/(0 Gauthier-Villars
308 D. M. GREENBERGER
that
electrodynamics
has been combined with the weak interaction. None- theless I believe that there is a unification withgravity
to behad, only
thatit will
proceed along
very different lines than have so far beeninvestigated.)
I believe that the very existence of the weak
equivalence principle
canonly
be understood in terms of theuncertainty principle,
and I will try here toexplain
this. It is easiest to describe these ideas in terms of asimple
extension of
special relativity,
which takes the mass moreseriously
as adynamical
variable.(The
details of thistheory
were worked out in Ref.[1 ].) Specifically,
thetheory
says that in fact we live not in a 4-dimensionalspacetime continuum,
but rather in a 5-dimensional one, and the fifth dimension is not ahidden, folded-up,
mathematicalfiction,
but is as realas the other four. It is the proper
time,
which should be treated as an inde-pendent degree
of freedom.Why?
Because like any truedegree
offreedom,
we can not
only arbitrarily
setit,
but we canarbitrarily
determine its rate ofchange.
One can
always (in principle)
erect aspherical
mass sheet around aregion
and this will affect thegravitational potential inside,
and hence the rate at which the proper time of aparticle
runs relative to coordinatetime
there,
withoutexerting
any forces in theregion
or otherwisedisrupting
the local
physics. (Thus
it isonly
the secondderivative,
the « acceleration » of the propertime,
that is affectedby
the laws ofmotion.)
The
dynamical
variableconjugate
to the proper time is the mass, andtogether they
enter the Hamiltonian of theproblem
in the same wayas x and p. This in turn determines the
equations
ofmotion,
whichyield
not
only x
as a function of thetime, t,
but also the propertime,
’r, as a func- tion of t(so
that this needs nolonger
to bepostulated
as part of the kine-matics). Also, just
as p canchange
as a function oftime,
so too can mchange
as a function of time. Thus the
theory
is a classicaltheory
ofchanging
mass, and
represents
a new andsimple
way to treatdecaying particles,
even
classically.
But for our purposes, we
merely
write as anillustration,
the Hamiltonian for afree,
stableparticle
Now in
ordinary physics, m
here isjust
a parameter, and there is noexpla-
’
nation for the symmetry between m and p.
However the
independent
variables here are x and ’t", to be determinedas
f(t) by solving
theequations
of motion. Theirconjugate
variables arep and m. For a free
particle
thechange of p
and m aregiven by
which
merely
says that momentum and 0 mass are ’ conserved 0here,
as isAnnales de l’Institut Henri Poincaré - Physique " theorique
expected.
But one sees thatjust
as one could have Hdepend
on x,producing
a force which
changes
p, one couldequally
as well make Hdepend
on ’t,which would
produce
a « force » that wouldchange
the mass of theparticle.
Yet even at this
simple level,
there is further information to begained,
for the other
equations
areThese
equations
determineand also
So this latter
equation
need not bepostulated separately,
as a property of themetric,
but followsdirectly
from the symmetry of thetheory.
In asense, when you were
taught relativity,
you weretaught only
half of thesubject. (Also,
in thistheory,
if theparticle decays,
this will affect the rate of passage of propertime.)
It is not necessary here to consider a more
complicated
system. Wemerely
have to
point
out that in thisdynamics,
besides the usualuncertainty principles
there is a new one
This last
equation
says that when one tries to measure the mass, or the propertime, directly,
there is a limit to the combined accuracy of the two, which is determinedby ordinary
quantum considerations.For
example,
let us assume we want to measure the mass,M,
of aheavy particle by gravitationally scattering
off of it alight particle
of knownmass, m, and
velocity v (see
thefigure).
The time of maximuminteraction, T,
isgiven by
The transverse momentum
picked
upby
thelight particle
will beVol. 49, n° 3-1988.
310 D. M. GREENBERGER
FIG. 1. - The Uncertainty Principle for Mass and Proper Time. If one measures the mass M of a heavy particle by scattering a light mass off of it gravitationally, then the angle 8 deter-
mines M. But the light particle exerts an unknown gravitational potential on M, to the ’
extent that b is unknown, so that ~, where T is the proper time read by a
clock located on M.
and the deflection will be
approximately (for
small9)
which
yields
a measure of M. But alsoThis
gives
both the mass and the accuracy to which it is known.However if there were a clock
sitting
onparticle M,
then even if we knewits time
exactly
before weperformed
theexperiment,
we would nolonger
know it
accurately
after theexperiment
has beenperformed.
This is becausewhile the
light particle passed by,
it exerted agravitational potential
onM,
which will affect the clock
reading.
If the distance b is known to within5b,
then
and one has
So the extent to which one can
accurately
measure the mass of theparticle
limits theknowledge
one has of its propertime,
and vice versa.One sees that this
only depends
on the usualuncertainty principle
betweenpy and y,
operating
on thelight particle,
and so is unavoidable.(For
furtherexamples
of theuncertainty principle,
see the second paper of ref.[1 ].)
There are any number of
examples
of thistype,
andthey
all show that aslong
as there is anuncertainty
relation between E and t, and p and x, then there must be one between m and r.It seems to me that this
uncertainty principle
between m and 03C4 leadsAnnales de l’Institut Henri Poincaré - Physique theorique
to an
important insight
into the nature of the weakequivalence principle,
because this
principle
states that in an externalgravitational
field the motion of aparticle
isindependent
of its mass, m.Thus,
in an externalgravitational
field one may know the mass verypoorly,
and yet stillaccurately
determine theposition
andvelocity
of theparticle.
In otherwords,
one can have(nonrelativistically)
So without
violating
theuncertainty principle,
one can know thetrajectory
of a
particle
veryaccurately, provided
one does not know its mass, because in this case, one can know thevelocity
withoutknowing
the momentum.But this
situation,
sototally
different from the usual quantum case whereone cannot know
trajectories,
ispossible only
because of theequivalence principle.
Forexample, Kepler’s
law states that for aparticle
in a circularorbit in a field with a
1/r potential, independently
ofthe mass of the
orbiting particle
m. Thus even if m is verypoorly
known(indeed,
because it ispoorly known),
one can measure the radius r accu-rately,
and determine theperiod
Taccurately
and therefore calculate thevelocity v (= 27rr/T)
and the proper time 1"(= ~/1 - r~/c~).
Therefore one very
important
index of theequivalence principle
is thatit allows one to work in the limit
Of course one can have intermediate cases where m is known to some
extent. But because the mass
drops
out of theequation
of motion in anexternal
gravity field,
itgives
rise to this veryspecial
property, that onecan determine
trajectories.
This is to be
strongly
contrasted with the case inordinary
quantumtheory,
where we
usually
know the massfairly accurately.
But in that situationwe know almost
nothing
about the propertime,
T2014a statement that needs to beexplained.
In our discussion we have assumed that one is
going
to measure m.This measurement
yields
theuncertainty
relation ~. Even if oneassumes that r was
accurately
known before theexperiment,
the uncertain- ties introducedby
the act of measurement willdestroy
thisknowledge
to the extent necessary. If one does not
perform
such anexperiment
butassumes, say, that one has an
electron,
then one must ask « Since when has this electron beenaround,
in order to start the clock for ’trunning ?
».If it is a stable
particle,
then Om =0,
and one has no idea when it wascreated,
and Ar=oo.Ifit is an unstableparticle
oneusually
writes~,
Vol. 49, n° 3-1988.
312 D. M. GREENBERGER
where ~E is the width of the
particle,
and Ot is the lifetime. But in fact this « DE » isreally
a massuncertainty. Similarly,
there is no real Ot asso-ciated with the
particle one
may know thelaboratory
timequite
accura-tely.
It is moreproperly
a statement about theuncertainty
of a clock locatedon the
particle,
that runs solong
before theparticle decays.
So many ofour usual lifetime-width
uncertainty
relations are moreappropriately represented
as mass-proper time relations.The above relates to the case where the mass is taken as «
given
», andone does not set about to determine it
experimentally.
Ifinstead,
aparticle decays,
and one wants to know whatproduct
onehas,
one can measureits mass, and
thereby
reset the propertime,
L, asdescribed,
butonly
to theextent
governed by
theuncertainty principle.
So L can benaturally
« set »by
the formation of theparticle,
or later « reset »by
aspecific experiment,
one
measuring
the mass, or rdirectly.
When the mass is taken as
given,
the usual case inquantum theory,
andone
applies
an electric(or
othernon-gravitational)
field to accelerate theparticle,
the usual restrictions on its orbitapply,
and one cannotaccurately
know its
trajectory,
which is needed to calculate r. As describedabove,
one does not know L because one does not know when to start
counting.
Over and above
that,
there is an extrauncertainty
because of ones lack ofknowledge
of thetrajectory.
Even in the case when oneignores
the initial lack ofknowledge
of L, and tries to determine the passage of r since the start of theexperiment,
one will be limitedby
the mass-proper time uncer-tainty relation,
where now Om will beDE/c2.
One can consider this expe- riment to be a new measure of the mass, and L, but after thesystem
loses itscoherence, Ar
will become infiniteagain.
So one can measure Ar in this manner in an external
non-gravitational field,
even if the initial mass isknown,
and theexperiment
thenyields
afinite Ar. However there will be a minimum
possible
Ar that one can obtainin such an
experiment.
One cannot make Ar =0,
because one cannotgain complete knowledge
of thetrajectory.
For a freeparticle,
the bestone can do is determined
by
the fact thatHere,
one does not know to, the creationtime,
atall,
as we have discussed.However,
if one wants to make a new determinationofr,
one will be limitedby
the fact that(assuming
for convenience a nonrelativisticparticle)
onecan not know both E and t
simultaneously,
eventhough
one would needto, in order to determine ’L, since
If m is taken as
known,
thenAnnales de l’Institut Henri Poincaré - Physique " theorique "
One can then take and minimize Ar with respect to At. One finds
approximately
and one sees that Ar grows in time. For a neutron, one would find that
over its lifetime there is a cumulated
uncertainty
1 cm,although
in
practice,
it wouldlikely
be agood
deallonger.
We are now in a
position
to see that theequivalence principle places
classical
electromagnetic
andgravitational
fields in aunique perspective.
In an external
gravitational field,
one need not know the mass atall,
and yet one canaccurately
measure thetrajectory
ofparticles,
and one canthen be in the situation where
On the other hand in an electric
field,
one often knows the massexactly,
and one can be in the situation where
Of course, in either case one may have
partial knowledge
of eithervariable,
but the distinction between the extreme cases comes about because of the
equivalence principle.
The two cases shown abovecorrespond
to twoopposite
extremes of theuncertainty relation,
much as in the case of wavevs.
particle knowledge
inordinary
quantumtheory.
In that case, whichwe are used to, one says that a quantum mechanical
object
issubject
toquantum
laws,
which in the two extremes reduce to the classical charac- terisation asparticle
motion or wave motion. Ingeneral,
neither is anadequate
characterisation.I would characterise the combined
gravitational
andelectromagnetic
field as a quantum
field,
which in the two extreme limits can be characterisedas a
gravitational field, subject
to theequivalence principle,
or else as anelectromagnetic
field. Ingeneral
it issomething
morecomplicated
thaneither.
(I
do not think the weak interactionchanges
thisscheme.)
So I would
expect
the search for a unified field toproduce
thisfully
quantum mechanical
entity, governed by
theuncertainty principle,
whichis neither electrical nor
gravitational
in nature. It isonly
in the extremelimits that we can characterise it as a classical field of one nature or the other.
Only
when thisgeneralized
field is understood will it becomepossible
to tackle the
problem
we refer to as « quantumgravity ».
I suspect that if present attempts toquantize
classicalgravity
were tosucceed,
it wouldbe a serious setback for
physics,
since everyone would rush to accept thetheory,
which wouldprobably
be wrong, and there would be no way to test it.In the
light
of theseremarks,
one can see that I believe that our presentVol. 49, n° 3-1988.
314 D. M. GREENBERGER
ideas as to the nature of
gravitational
andelectromagnetic
fields are stillhopelessly naive,
from thestandpoint
of a « final »theory.
I would not fora moment
question
the mathematicalsophistication
of presenttheories,
but I believe that there are still many
physical insights
that we have notmade use of.
Certainly
theequivalence principle
still has much to teach us.ACKNOWLEDGMENTS
I would like to thank Prof. Herbert Walther and the Max Planck Institut fur
Quantenoptik
for the wonderfulhospitality
afforded meduring
mystay
here,
and also the Alexander von HumboldtStiftung
for their supportduring
thisperiod,
as well as theCity College
of theCity University
ofNew York.
[1] D. M. GREENBERGER, J. Math. Physics, t. 11, 1970, p. 2329 ; t. 11, 1970, p. 2341 ; t. 15, 1974, p. 395 ; t. 15, 1974, p. 406.
(Manuscrit reçu le 10 juillet 1988)
Annales de Henri Poincaré - Physique theorique